Using laboratory experiments within a New Keynesian framework, we explore the interaction between the formation of inflation expectations and monetary policy design. The central question in this paper is how to design monetary policy when expectations formation is not perfectly rational. Instrumental rules that use actual rather than forecasted inflation produce lower inflation variability and reduce expectational cycles. A forward-looking Taylor rule where a reaction coefficient equals 4 produces lower inflation variability than rules with reaction coefficients of 1.5 and 1.35. Inflation variability produced with the latter two rules is not significantly different. Moreover, the forecasting rules chosen by subjects appear to vary systematically with the policy regime, with destabilizing mechanisms chosen more often when inflation control is weaker.
